Question 14
Domain 2: Fairness, Bias, and Societal ImpactA team is building an AI system and wants to reduce bias throughout the project rather than only at the end. Which approach best aligns with bias-mitigation controls across the AI lifecycle?
Correct answer: C
Explanation
Bias mitigation should be addressed as a continuous lifecycle activity, with controls applied during data collection, preprocessing, model design, testing, and ongoing monitoring. Limiting controls to a single stage leaves other sources of bias unaddressed. — Source material: Propose controls during data collection, preprocessing, model design, testing, and monitoring.
Why each option is right or wrong
A. Apply bias controls only during data collection, because later stages mainly affect accuracy rather than fairness.
Controls are proposed during data collection, preprocessing, model design, testing, and monitoring.
B. Focus bias controls on preprocessing and testing, since those are the main stages where unfair outcomes can be detected and corrected.
Controls span all listed lifecycle stages, not only preprocessing and testing.
C. Implement bias controls during data collection, preprocessing, model design, testing, and ongoing monitoring.
The source material explicitly identifies controls during data collection, preprocessing, model design, testing, and monitoring. Because the team wants mitigation across the lifecycle, the only choice that covers all named stages is this one.
D. Reserve bias controls for post-deployment monitoring, because real-world use is the first point where bias can be meaningfully assessed.
Monitoring is only one listed stage; earlier controls are also required.